The Syntactic Licensing of Ellipsis -- Editorial page -- Title page -- LCC data -- Table of contents -- Preface -- Abbreviations and formats used in examples and glosses -- What is ellipsis? -- 1.1 A mismatch between sound and meaning -- 1.2 Different views on ellipsis -- 1.2.1 Nonstructural approaches -- 1.2.2 Structural approaches -- 1.2.2.1 Null proforms/LF copy -- 1.2.2.2 PF-deletion -- 1.2.3 Summary -- 1.3 Restrictions on ellipsis -- 1.3.1 Recoverability -- 1.3.2 Licensing -- 1.4 Overview of this work -- Dutch modal complement ellipsis -- 2.1 Modal complements in Dutch -- 2.1.1 Epistemic, deontic, and dynamic modals -- 2.1.2 Are modals raising or control? -- 2.1.2.1 Diagnostic tests for the raising/control distinction -- 2.1.2.2 Both deontic and epistemic modals are raising verbs -- 2.1.3 The categorial status of Dutch modals and their complement -- 2.1.3.1 The modal complement -- 2.1.3.1.1 The complement of a modal contains a vP layer. Given that epistemic and deontic modals are raising verbs, the base-generation position of the subject must be below the modal. In other words, the complement of an epistemic or deontic modal must be at least a vP, where the external argument is introduced in [Spec,vP] and little v assigns structural Case to the object (Chomsky 1995). Dynamic willen 'want' and durven 'dare', which are control verbs, also select at least a vP, because there -- 2.1.3.1.2 The complement of a modal contains tense. There is a whole body of literature on the presence or absence of tense in infinitival complements (see Karttunen 1971 -- Stowell 1981, 1982 -- Pesetsky 1992 -- Guasti 1993 -- Rizzi 1993 -- Haegeman 1995 -- Martin 1996 -- Bošković 1995, 1996 -- Boivin 1998.

Felser 1998 and Wurmbrand 2003). In my discussion of modal complements I add to this debate, claiming that Dutch modal complements contain tense and hence are TPs. -- 2.1.3.1.3 The complement of a modal is not a CP. In the previous section we have established that modals select an infinitival complement that contains at least a tense projection. The next question to be asked is whether the modal complement is as big as CP, a full clause. Several arguments indicate that the answer to this question is negative (see also Barbiers 2005). A first argument for this claim is that there is never an overt complementizer in infinitival complements of modals (cf. (36)a, -- 2.1.3.2 The categorial status of Dutch modal verbs -- 2.1.3.2.1 Dutch modals are not inflectional heads. Although much has been said about the category of English modals (Chomsky 1957 -- Jackendoff 1972, 1977 -- Fiengo 1974 -- Akmajian, Steele & Wasow 1979 -- Palmer 1983, 1986, 1990, 2001 -- Bobaljik 1995 -- Bobaljik and Thráinsson 1998), Dutch modals have received relatively little discussion. I compare Dutch modals to English ones, showing that they cannot be considered of the same categorial status. -- 2.1.3.2.2 Dutch modals are not auxiliaries. I have argued that Dutch modals, as opposed to English modals, are not inflectional heads. The next option to consider is whether they are auxiliaries (see also Abraham 2002 for German modals). -- 2.1.3.2.3 Dutch modals are Mod/V heads. In the literature modal verbs in Dutch and German have been considered the head of their own functional projection ModP, and not lexical V heads (see Cinque 1999, Wurmbrand 2003, Barbiers 2005, among others). Overall, however, Dutch modals do not behave significantly different from other verbs that can select a (tensed) infinitival complement, such as besluiten 'decide' or leren 'learn'. -- 2.1.4 Summary.

2.2 Dutch modal complement ellipsis: Properties -- 2.2.1 MCE is only allowed with root modals -- 2.2.2 MCE affects a complete constituent -- 2.2.3 Extraction -- 2.2.3.1 Subject extraction -- 2.2.3.2 Object extraction -- 2.2.3.2.1 Object scrambling. Dutch displays a form of word reordering called object scrambling. Pronominal objects have to move out of the verb phrase across negation. The same holds for definite full DP objects, although scrambling is preferred rather than obligatory in this case. This is illustrated in the following examples in (82) where the direct objects je 'you', het 'it' and dat boek 'that boek' and the indirect object hem 'him' cannot follow sentential negation and certain adverbs. -- 2.2.3.2.2 Wh-object extraction. Next, I consider the less clear-cut cases. I first discuss wh-extraction out of the MCE ellipsis site and then turn to topicalization. Recall that the sentence in (81), repeated here as (87), is ungrammatical as a result of the wh-object extraction. -- 2.2.3.2.3 Topicalization. A third kind of object extraction out of the ellipsis site, next to scrambling and wh-movement, is topicalization. Dagnac (2007) argues that in French MCE topicalization is possible. In the examples in (94)a,b the direct object is contrasted with the object in the antecedent and has undergone movement to the left periphery of the clause -- and in (94)c it is the PP indirect object that is topicalized. -- 2.2.3.3 Adjunct extraction -- 2.2.3.4 Summary -- 2.2.4 There-sentences and MCE -- 2.2.5 MCE blocks the IPP effect -- 2.2.6 Restrictions on the position of the antecedent -- 2.2.6.1 Embedded antecedent, embedded ellipsis site -- 2.2.6.2 MCE applies across utterance boundaries -- 2.2.6.3 MCE allows backward anaphora -- 2.2.7 Form mismatches between antecedent and ellipsis site -- 2.2.8 Summary -- Ellipsis licensing -- 3.1 Licensing via Agree.

3.1.1 The ellipsis licensing head -- 3.1.2 Merchant (2001) -- 3.1.3 Material between licensor and ellipsis site -- 3.1.4 Ellipsis licensing via Agree -- 3.1.5 Summary -- 3.2 Derivational ellipsis -- 3.2.1 The timing of ellipsis -- 3.2.2 Returning to the extraction puzzle -- 3.2.3 Derivational ellipsis -- 3.2.3.1 Limited extraction -- 3.2.3.2 Limited LF movement -- 3.2.3.3 Summary -- 3.2.4 Ellipsis versus phases -- 3.2.4.1 Ellipsis and phases (Gengel 2007a) -- 3.2.4.2 Differences between ellipsis and phases -- 3.2.5 Interaction between ellipsis and phases: Predictions -- 3.2.5.1 An intervening phase head -- 3.2.5.2 No intervening phase head -- 3.3 The analysis of MCE -- 3.4 Explaining the properties of MCE -- 3.4.1 Extraction -- 3.4.1.1 Subject extraction with transitive verbs -- 3.4.1.2 Extraction of a derived subject -- 3.4.1.3 Extraction of a wh-object -- 3.4.1.4 Object scrambling -- 3.4.1.5 Adjuncts -- 3.4.1.6 Antecedent-Contained Deletion -- 3.4.1.7 Summary -- 3.4.2 There-expletives and MCE -- 3.4.3 MCE blocks the IPP effect -- 3.5 The locality restriction on Agree -- 3.6 Summary -- Extending the analysis to other ellipses -- 4.1 Sluicing -- 4.1.1 The licensing head and ellipsis site for sluicing -- 4.1.2 Applying the analysis to sluicing -- 4.1.2.1 An [E]-feature for sluicing -- 4.1.2.2 Predictions for extraction -- 4.1.3 Summary -- 4.2 English VP ellipsis -- 4.2.1 The licensing head of English VP ellipsis -- 4.2.2 The VP ellipsis site -- 4.2.3 Applying the analysis to VPE -- 4.2.4 Accounting for the properties of VPE -- 4.2.4.1 Extraction -- 4.2.4.2 VPE and there-expletives -- 4.2.5 Summary -- 4.3 Pseudogapping -- 4.3.1 The licensor and the ellipsis site of pseudogapping -- 4.3.1.1 The licensing head -- 4.3.1.2 The pseudogapping ellipsis site -- 4.3.2 The movement operation in pseudogapping -- 4.3.2.1 Heavy NP Shift -- 4.3.2.2 Object Shift.

4.3.2.3 Focus movement (Gengel 2007b) -- 4.3.3 Applying the analysis to pseudogapping -- 4.3.3.1 An [E]-feature for pseudogapping -- 4.3.3.2 Predictions for extraction -- 4.3.4 Summary -- 4.4 British English do -- 4.4.1 The properties of British English do -- 4.4.2 The analysis of British English do -- 4.4.2.1 The licensor of British English do -- 4.4.2.2 The ellipsis site of BE do -- 4.4.2.3 An [E]-feature for British English do -- 4.4.3 Accounting for the extraction data -- 4.4.4 Baltin (2007) -- 4.4.5 Summary -- 4.5 Summary -- Conclusion and issues for further research -- References -- Index -- The series Linguistik Aktuell/Linguistics Today.